Kimi review

Kimi, from Moonshot AI, is the outsider here and the most agent focused. Its membership plans are built around agent credits: Plus costs $15 per month on annual billing, or $19 monthly, and adds more credits, agents that work with documents, spreadsheets and slides, Deep Research, website deployment, multi-tasking agents and scheduled tasks that run on their own and push results. Pro at $31 per month on annual billing doubles agent credits and adds a goal mode where agents work until a task is done. Kimi also lists an extra long chat capacity of up to 1M tokens, which helps with very large documents. The drawbacks are maturity and clarity. The pricing page warns that K3.1 Pro requests may queue or fail at launch, and plans described in credits and concurrent tasks are harder to compare than the simple usage tiers of ChatGPT or Claude. It is worth a look for agent-heavy research, less so as a first assistant.

Who it is for

Power users who want agent tasks, deep research and very long chats at a mid-range price.

Price

Free plan; Plus $15 per month billed annually or $19 monthly; Pro $31 per month billed annually.

$ 15 / month $19 Billed annually $180
Quoted from the vendor pricing page on November 9, 2026.

Key features

• Agents that work with documents, spreadsheets and slides
• Deep Research on paid plans
• Scheduled tasks that run automatically and push results
• Extra long chat capacity of up to 1M tokens
• Website deployment with database support

Strengths and weaknesses

• Strength: Plus is $15 per month on annual billing
• Strength: Scheduled tasks are included from Plus
• Strength: Long chat capacity of up to 1M tokens
• Weakness: Moonshot warns that K3.1 Pro requests may queue or fail at launch
• Weakness: Plan details lean on agent credits, which are harder to compare
• Weakness: Plus runs 2 agent tasks at once; 4 concurrent tasks need Max

Key facts

• Free plan: Yes
• Paid tiers: Plus, Pro, Max, Ultra
• Long chat capacity: Up to 1M tokens

Where it ranks

Kimi is No. 8 in our Best AI Chatbots verdict.